Meta rolls out global 'Plus' subscriptions for social platforms

🔑 Enhanced Key Takeaways

•Meta's new 'Plus' subscriptions for Facebook and Instagram, priced at $3.99/month, offer features like aggregate Story rewatch counts, unlimited audience lists, custom app icons, and customizable profile fonts, while WhatsApp Plus ($2.99/month) focuses on personalization through app themes, custom ringtones, and premium stickers.
•The company is consolidating its various subscription offerings under a new 'Meta One' brand, which includes consumer-facing 'Plus' tiers, professional packages for creators and businesses (Meta One Essential at $14.99/month and Meta One Advanced at $49.99/month), and AI-specific subscriptions.
•Meta AI will offer paid tiers, Meta One Plus ($7.99/month) and Meta One Premium ($19.99/month), providing enhanced capabilities such as higher compute queries, deeper reasoning through a 'thinking mode,' and expanded image and video generation, while a free, ad-supported tier with limited access to the Llama 4 Turbo model will remain available.
•Meta's substantial AI infrastructure investments are projected to reach $125-$145 billion in capital expenditures for 2026, a significant increase from $69.7 billion in 2025, with a multi-year commitment of $600 billion over three years for US AI and data center expansion.

📊 Competitor Analysis▸ Show

Competitor Analysis: Social & AI Subscriptions

Feature/Service	Meta (Plus/Meta One)	X (Premium)	Snapchat+	ChatGPT (Plus/Pro)	Google AI (Gemini Pro/Ultra)	Claude (Pro/Max)
Social App Pricing	FB/IG Plus: $3.99/month; WhatsApp Plus: $2.99/month; Meta One Essential (Pro): $14.99/month; Meta One Advanced (Pro): $49.99/month	Premium: $8/month (web), $11/month (iOS/Android); Premium+: $16/month (web)	$3.99/month	N/A	N/A	N/A
Social App Features	Story rewatch counts, unlimited audience lists, custom app icons, profile fonts, app themes, custom ringtones, premium stickers, Verified badge, impersonation protection, enhanced linksheet, higher search placement	Edit posts, longer posts, fewer ads, creator monetization, verified badge, custom app icons, NFT profile pictures, longer video uploads	Exclusive app icons, badge, Story rewatches, Story boosts, custom chat backgrounds, priority Story replies, post view stats	N/A	N/A	N/A
AI Service Pricing	Meta One Plus: $7.99/month; Meta One Premium: $19.99/month	SuperGrok: $30/month; SuperGrok Heavy: $300/month	N/A	Plus: $20/month; Pro: $100-$200/month	AI Pro: $19.99/month; AI Ultra: $100-$200/month	Pro: $20/month; Max: $100-$200/month
AI Service Features	Higher compute queries, deeper reasoning ('thinking mode'), expanded image/video generation, Llama 4 Deep Think model access, priority queues, 90-day memory	Increased access to Grok 3/4, extended token limits (128K), priority voice access, Imagine image model, AI companions	N/A	GPT-4o/5.5 access, image generation, browsing, data analysis, advanced voice/video/screensharing, agent mode, deep research	Gemini 3 Pro/3.5 Flash, Deep Research, 2TB storage, Workspace integration, Veo 2 video generation, YouTube Premium, Google Antigravity	Claude models access, Research, extended thinking, higher usage limits, Claude in PowerPoint (research preview), Comet browser agent
AI Model	Llama 4 (Scout, Maverick, Deep Think)	Grok 3, Grok 4	N/A	GPT-4o, GPT-5.5	Gemini 3 Pro, Gemini 3.5 Flash	Claude models
Benchmarks	Llama 3.1 405B achieves GPT-4o parity on several reasoning benchmarks; Llama 4 Scout is a top multimodal model in its class, outperforming Gemma 3, Gemini 2.0 Flash-Lite, and Mistral 3.1	N/A	N/A	N/A	N/A	N/A

🛠️ Technical Deep Dive

Llama Model Architecture: Meta's Llama family of large language models (LLMs) includes Llama 4, which represents a significant architectural shift. Llama 4 models, such as Scout (109B total parameters with 17B active) and Maverick (400B total parameters with 17B active), utilize a Mixture-of-Experts (MoE) architecture. This means that while all parameters are stored, only a subset are activated per token during inference, enhancing efficiency.
Context Window Expansion: Llama 4 Scout dramatically increases the supported context length to an industry-leading 10 million tokens, a substantial leap from Llama 3's 128K tokens. This enables advanced capabilities like multi-document summarization and reasoning over vast codebases.
Multimodality: Llama 4 was designed with native multimodality from the ground up, allowing for more personalized multimodal experiences.
AI Infrastructure Scaling: Meta's AI network infrastructure is built around 'AI zones' using a two-tier Clos topology. The company scaled from a 24K GPU cluster for Llama 3 training to a 100K+ GPU multi-building cluster for Llama 4, addressing challenges like network congestion, priority flow control, and buffer management.
Hardware Investment: Meta has committed to deploying Nvidia's Blackwell and upcoming Rubin GPUs, along with standalone Grace CPUs and Spectrum-X networking hardware, in a multi-year partnership estimated at tens of billions of dollars.
Custom Chips: Before 2022, Meta used CPUs and in-house custom chips, including MTIA v1, designed for content recommendation algorithms, fabricated on TSMC's 7 nm process technology.

🔮 Future ImplicationsAI analysis grounded in cited sources

Meta will significantly reduce its reliance on advertising revenue.

The global rollout of diverse subscription tiers across its core social platforms and AI services demonstrates a concerted effort to build substantial recurring revenue streams, diversifying its business model beyond its traditional ad-centric approach.

The aggressive AI infrastructure investments will accelerate Meta's pursuit of Artificial General Intelligence (AGI).

Meta's commitment of $125-$145 billion in capital expenditures for 2026 and a multi-year $600 billion investment in AI infrastructure are explicitly aimed at scaling AI computing capacity to develop and deploy advanced models, with a stated goal of achieving AGI by 2027.

The multi-tiered subscription strategy risks creating a 'pay-to-play' perception among users for advanced features.

By offering exclusive features and enhanced capabilities behind paywalls for both social apps and AI, Meta may inadvertently segment its user base and potentially alienate users accustomed to free access to core functionalities.

⏳ Timeline

2013

Facebook Artificial Intelligence Research (FAIR) founded.

2021

Facebook Inc. rebranded to Meta Platforms Inc., emphasizing the metaverse.

2023-02

Llama 1, Meta's first large language model, released to researchers.

2023-07

Llama 2 released with a more permissive commercial license.

2025-04

Llama 4 released, introducing Mixture-of-Experts architecture and native multimodality.

2026-05

Meta announces $125-$145 billion in AI capital expenditures for 2026 and rolls out global 'Plus' subscriptions for social platforms and tests Meta AI subscriptions.

Meta rolls out global 'Plus' subscriptions for social platforms

⚡ 30-Second TL;DR

Key Points

🧠 Deep Insight

🔑 Enhanced Key Takeaways

Competitor Analysis: Social & AI Subscriptions

🛠️ Technical Deep Dive

🔮 Future ImplicationsAI analysis grounded in cited sources

⏳ Timeline

📎 Sources (32)

👉Related Updates

The vertical video takeover is here